Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Chapter 2
10
Today, connection yield strength is related to wood crushing and bolt bending strength,
and directly determines joint capacity and yield mode. The idea was first introduced by Johansen
(1949) who linked the elastic bending capacity of the fastener to joint strength. In 1957 Meyer
employed the full plastic bending capacity to determine joint strength, which is still used today
(Figure 2.1).
Fastener in bending
elastic
plastic
max
max
4r
3
Fresultant = max
Cross section
Fresultant = max
r2
2
4r
3
r2
2
Figure 2.1: Illustration of elastic and plastic bending capacity of the fastener
Elastic bending moment:
M el = max, elastic
d3
32
(2.1)
M pl = max, plastic
Mmax
max
d
r
F
2r2 4r
d3
= max
2
3
6
1
Used in this work to describe both 5 percent offset moment and maximum moment.
(2.2)
Unit
Nm
N/m2
m
m
N
Chapter 2
11
Fastener yield strength is determined experimentally. In the U.S., the fastener is bent in
three-point loading, and the load at 5 percent fastener diameter (used to compute Mmax) is
converted into bending yield strength using the equation
max =
6 M max
d3
(2.3)
In Europe, bending yield strength is also determined by Equation (2.3), but input
parameters are acquired differently.
maximum bending moment and the moment at a deformation of 45 under three-point loading.
The methodology is less conservative than the U.S. approach, as the 5 percent offset bending
yield strength falls in-between proportional limit strength and ultimate bending strength of the
fastener. Harding and Fowkes (1984) originally suggested the 5 percent diameter offset load
resulting in the adoption by the 1991 edition of the NDS. The main disadvantage of the offset
approach is that it neither uses proportional limit nor capacity as a reference point. Moreover,
attributed to joint settlement effects as loading commences, the fitting of an initial stiffness line is
frequently ambiguous and judgmental. In addition, the offset approach is difficult to program if
the analysis is conducted by a computer. A capacity based approach, on the other hand, exercised
with appropriate adjustment factors to account for variability is relatively easy to program, clearly
defined and promotes best material utilization.
Neither method accounts for strain hardening of the fastener. Due to hardening, the
plastic bending resistance is not constant with increasing deformation. It follows that the plastic
bending stress distribution is not uniform, resulting in altered fastener yield strength (Figure 2.2).
For joints with slender dowels that undergo large deformations when loaded to failure, the true
fastener yield strength is higher due to hardening. Mischler (1998) recommended that the true
bending resistance of the dowel at the corresponding deformation be used as input in theoretical
models.
Chapter 2
12
Materials reviewed the standard and now recommends the full-hole test for specimens with high
tendency to split (ASTM 1998). The full-hole test has the disadvantage that, because of the way
the fastener is loaded, it is almost impossible to avoid fastener bending, which implies the
embedding stress is not constant over the whole length of the dowel.
F
Europe
F
U.S.
Chapter 2
13
Load
longitudinal splitting
Slip
Load
Crack
Slip
lateral splitting starts
Figure 2.4: Typical load-slip curve of dowel embedment tests parallel and
perpendicular to grain
Wilkinson (1991) derived the following empirical formulation relating bolt diameter and
specific gravity to embedment strength for perpendicular and parallel to grain loading
(2.4)
Chapter 2
14
1.45 0.5
Fe( perp ) = 6,620 G12
d
Fe
d
G 12
G
embedment strength
fastener diameter
specific gravity based on ovendry weight and
volume at 12 percent moisture content
specific gravity based on ovendry weight and
volume
(2.5)
Unit
psi
inch
The researcher did not find a significant effect of diameter on bearing strength parallel to
the grain. Similar to fastener yield strength, embedment strength was determined as the load at 5
percent fastener diameter. In other words, Wilkinson derived Equations (2.4) and (2.5) based on
the 5 percent offset load rather than capacity. Equations (2.4) and (2.5) include specific gravity
based on ovendry weight and volume at 12 percent moisture content. Since specific gravity is
more frequently reported at ovendry weight and volume, Equations (2.4) and (2.5) were changed
to
(2.6)
(2.7)
and were adopted by the 1991 edition of the NDS. To convert the results of Equations (2.6) and
(2.7) to metric units and timber density at 20C/65% r.h., the following equations may be used.
(2.8)
(2.9)
Earlier, Whale and Smith (1986) conducted embedment tests in an effort to provide an
empirical formulation for the first draft of Eurocode 5. As opposed to Wilkinson, Whale and
Smith found a significant influence of bolt diameter on embedment strength parallel to the grain.
Fe =
(2.10)
k = 1.35 + 0.015 d
(2.11)
Chapter 2
15
and softwoods
k = 0.9 + 0.015 d
Fe
Fe
d
G
12
G
12,12
embedment strength
embedment strength
fastener diameter
fastener diameter
specific gravity based on ovendry weight and
volume at 12 percent moisture content
specific gravity based on ovendry weight and
volume
wood density based on weight and volume at 12
percent moisture content (20C/65% R.H.)
angle between direction of load and direction of
grain
(2.12)
Unit
N/mm2
psi
mm
inch
Kg/m3
Figures 2.5 and 2.6 portray European and converted US embedment strength models for
the purpose of comparison. In general, as reflected by the graphs, embedment strength increases
under fasteners with smaller diameter (except for Eqs. (2.8) and (2.9) where embedment strength
is not influenced by diameter). This is attributed to the fact that for larger diameter fasteners
splitting occurs due to perpendicular to grain tension stresses soon after proportional limit load is
reached, and the fact that the embedment stress at proportional limit is not significantly different
from the embedment stress at failure (Jorissen 1998). It is interesting to note that Whale and
Smith obtained embedment strength at the maximum load in contrast to Wilkinson. Yet, for
parallel to grain (except for small diameter bolts), the US model yields higher embedment
strengths for a given species. Parallel to grain both models describe a linear relationship, whereas
for perpendicular to grain, Wilkinson found an exponential correlation between fastener strength,
bolt diameter and embedment strength as opposed to the European researchers, but did not find a
significant difference between hardwoods and softwoods. Again, the fact that Wilkinson used the
5 percent offset load to determine embedment strength may explain why his model yields smaller
values at larger fastener diameters, but gives larger values at smaller diameters when compared to
the European model for softwoods. It should further be noted that Wilkinson used the half-hole
test with uniform pressure application along the dowel as described above, while Whale and
Smith assessed embedment strength employing the full-hole test, which was later included in
European test standards. The full-hole tests produces both dowel bearing and dowel bending
when small diameter fasteners are used.
Chapter 2
16
20
30
50
40
30
20
700
600
Tim ber d
e nsit y at
500
400
20 C/65%
r
300
.h. (kg/m 3
)
Europe (hardwoods)
10
(m m)
20
30
US
40
20
Europe (softwoods)
0
500
400
300
600
700
3
kg/m )
(
.
h
.
5% r
0 C/ 6
2
t
a
y
sit
er de n
Timb
Bolt diam et er
Chapter 2
17
t
d
(2.13)
where:
t = member thickness
d = fastener diameter
Obviously this definition implies that one obtains more than one aspect ratio for joints
with different member thicknesses. Early research (Trayer 1932) identified that fastener aspect
ratio is the principal factor determining bearing stress. It is well established that constant aspect
ratio produces constant average bearing stress in wood as long as the fastener diameter does not
fall within growth ring width.
Fastener aspect ratio determines to what degree fastener yield strength and embedment
strength influence joint behavior. It controls the yield and failure modes of the joint. Joints
containing rigid fasteners (small aspect ratios) tend to exhibit brittle failure modes on account of
wood splitting, whereas slender fasteners bend and develop plastic hinges. At decreasing aspect
ratios, more wood yielding and less fastener bending is involved. Figure 2.7 depicts the relation
between proportional limit stress and fastener slenderness as found by Trayer (1932) who
assumed uniform stress distribution beneath the bolt2. His tests on joints in double shear showed
that for rigid fasteners with aspect ratios of smaller or equal to two, joint strength is fully
governed by embedment strength as no appreciable bending of the fastener takes place. Hence,
with no fastener bending involved, the proportional limit stress remains constant for changing
aspect ratios, neglecting volumetric strength variations of wood.
Trayer obtained the proportional limit stress by dividing the proportional limit load by the projected area
of the fastener
18
Chapter 2
n +1
= n
n
n +1
(2.14)
The subscript n+1 denotes the next larger aspect ratio with the corresponding proportional limit
stress. As a result, for fasteners with aspect ratio greater than s the proportional limit load of the
joint remains constant for a certain fastener diameter (Figure 2.7). This observation was later
confirmed by Johansens Yield Theory (Johansen 1949). In other words, Trayer discovered what
is known today as Mode IV yield. For double shear connections Trayer determined s to be about
5 .
decreasing edge distance. Studies of stress distributions around pin loaded holes have been
mainly concerned with the magnitude and direction of stresses at the immediate pin/hole interface
and have assumed that the medium around the fastener is of infinite dimension. Yet, stress
distribution away from the hole is significantly influenced by the direction of loading in a finite
member. Tensile loading causes the stress lines to run around the bolt resulting in tensile stresses
Chapter 2
19
Capacity is
Figure 2.8: Stress lines due to tensile or compressive loading (after Jorissen
1998)
2.2.5 Friction
Bolted connections with nuts drawn tight develop significant friction between the
abutting surfaces, which increases the capacity of the joint. Nonetheless, due to the rheological
characteristics of wood, the compression force achieved by tightening the nuts is quickly lost. In
addition, possible moisture related dimensional changes of the wood members turn surface
friction between members into a rather variable element. As a result, some researchers have
tested joints with bolts drawn fingertight or left loose (Trayer 1932, Johansen 1949).
Considerable friction may, however, develop past the proportional limit when the bolts bend and
draw the members against each other.
Chapter 2
20
members do not provide much resistance to bolt-end rotation despite the restraint imposed by
head, nut and washers. However, analysis of test data reported in the literature by McLain and
Thangjitham (1983) lead to the discovery that in single shear joints containing thicker side
members (76 mm) bolt end fixities tend to reduce the likelihood of end rotation.
On the other hand, end fixity may introduce tensile forces in the bolt and may cause force
components acting against the applied force (McLain and Thangjitham 1983).
At higher
displacements, the bolt or part of the bolt rotates and the inclined tensile force causes a force
component in the direction of loading. Furthermore as a consequence of axial bolt tension, the
normal force between the two joint members increases leading to a frictional force component in
loading direction.
Since bolted joints must contain washers under head and nut by code regulation, the
effects of end fixity should be included in any analytical or numerical model attempting to
accurately replicate joint performance.
Chapter 2
21
embedment strength. Rodd (1988) showed that interface friction indeed influences embedment
strength in wood in a way that embedment strength decreases with decreasing interface friction.
Figure 2.9 illustrates a pin-loaded hole with interface friction between fastener and wood.
Friction causes crushing of fibers underneath the fastener rather than the bending of fibers around
the bolt, which takes place at locations where friction is overcome (outside the compression
region as shown in Figure 2.9). This triggers splitting at two locations underneath the bolt. The
figure also exhibits the case where interface friction is reduced to zero and the splitting force, Fs,
acting perpendicular to the grain, becomes significantly larger. From the foregoing, it can be
concluded that a higher friction coefficient, , increases the compression region angle, , and the
wood is less likely to split (Rodd 1988, Patton-Mallory 1996, Jorissen 1998). The smaller , the
closer together the two cracks move, and the lateral displacement of the fibers is increased, which
produces a higher wedging effect resulting in wood splitting at lower loads. At zero friction, only
one crack would theoretically form in the center underneath the fastener.
Interface Friction (f)
No Friction (nf)
no-contact region
Fy / 2
Fy / 2
Fr,nf
Fr,f
Fs,f
Fs,nf
F,f
compression
region,
contact region
Chapter 2
22
Jorissen (1998) confirmed the formation of two cracks originating at the fastener hole
through tests on double shear joints containing rigid bolts loaded parallel to the grain. He
approximated the location of the cracks by assuming that crack formation commences at the point
where the frictional force, F, equals the tangential force component Ft, of the lateral force Fy
(Figure 2.10).
2 b = d sin
Fy
Ft
F
d
(2.15)
Units
N
N
N
mm
degrees
lateral force
tangential force component
frictional force
fastener diameter
angle of friction
Fy
Fy / 2
Fr
Ft
F < Ft
F < Ft
2
cracks
h
F > Ft
Chapter 2
23
2.3.1.2 Clearance
Fastener-hole clearance not only influences the overall stress magnitude around the hole,
but also affects the location and direction of stress maxima, because it decreases the contact
region and increases stress concentration (Hyer et al. 1987, Eriksson 1986). Furthermore, it is
suspected that hole clearance explains some of the uneven load distribution observed on multiple
bolt joints.
2.3.1.3 Stress Distribution
Knowledge of stress distributions is essential to predict failure accurately. Yet, the true
stress distribution around a pin-loaded hole in orthotropic materials is very complex. Under the
assumption of plain stress, the analytical derivation to determine the true stress distribution in an
orthotropic plate containing a hole, loaded in tension or compression was established by
Lekhnitskii (1968). His intricate functions have been incorporated into many theoretical works.
The approximate stress distribution along the two principal axes of a pin-loaded plate with no
clearance is shown in Figure 2.11. As a result of stress concentrations at the hole boundary, in
the directions perpendicular and parallel to the load, stresses parallel to the grain decrease rapidly
with increasing distance to the hole, both in transverse and longitudinal direction.
Py
y
Py
y
Py
x
Py
Figure 2.11: Stress distribution along the x-and y-axis of a pin-loaded plate.
A major contribution was made by de Jong (1977, 1982) who derived stress distributions
around a pin-loaded hole in an infinite orthotropic plate based on Lekhnitskiis work. De Jongs
late works include the effects of friction, but assume a rigid pin and no initial clearance between
pin and hole. A good synopsis of the calculations was given by Ireman et al. (1993). For further
information, the reader is referred to Ireman et al. (1993), de Jong (1977/82), and Lekhnitskii
(1968).
Chapter 2
24
In 1977, De Jong presented the stress distribution parallel and perpendicular to the fibers
around a pin-loaded hole in a unidirectional composite plate whose behavior in the elastic region
would be similar to a wooden plate. The analysis precluded friction, and a tight fit was assumed.
On grounds that stress concentrations next to a hole reduce quickly with increasing distance, De
Jong attempted to approximate a joint of finite width but infinite length by superposition of two
loading scenarios (Figure 2.12).
distributions (normalized based on average bearing stress) in the principal x-and y-directions of a
unidirectional carbon fiber reinforced plate of finite width but infinite length (Figure 2.13 No
Friction). The results can be extended to a wooden plate containing an infinitely rigid dowel,
which is loaded in the elastic range and moves in the y-direction only (Mode I).
Fy
Fy
2s
2s
Fy
s > 2.5 d
s > 2.5 d
Fy
Fy
2s
2s
Fy
s > 2.5 d
Fy
s
Figure 2.12: Approximation for a plate with finite width through superposition of
two loading scenarios in infinite plates (after De Jong 1977).
Chapter 2
No Friction
25
Friction
(angle of friction: = 22)
(x)
(x)
0 30
60
90
120 150180
0 30
60
90
120 150180
(y)
(y)
load direction
y
load direction
y
Chapter 2
26
= (1 2 cos 2)
(2.16)
3
r
y
x
Sc =
max 3 1
=
n
+ 0 .3
(2.17)
Units
Sc
N/m2
Chapter 2
27
2.3.1.5 Comments
While these studies are extremely useful because they present a closed-form solution and
give some insight about stress distributions and failure modes, their application to timber joints is
limited. Only Mode I-type deformations in the elastic range can be approximated with this
approach, as other deformation modes would violate the plain-stress assumption. Furthermore,
contemporary design practices for timber joints allow for plastic deformation of the joint. Yet,
exact stress prediction beyond the elastic limit is extremely complex without numerical
approximation.
assumes that the foundation acts both in tension and compression. Hence, the beam is always in
contact with the foundation and does not lift off.
According to Hetnyi (1946), the deflection curve of a beam on a Winkler foundation
between concentrated transverse loading forces can be described by the differential equation
EI
d4y
dx
d 2M
dx
dQ
= k y
dx
(2.18)
(2.19)
where
1
4
k
=
4 E I
(2.20)
Chapter 2
E
I
k
x
y
Ci
28
Units
N/m2
m4
N/m2
m
m
1/m
Using the Winkler-foundation assumption, Bhat and Xirouchakis (1986) found an exact
solution for the deflection profile of an elastic-plastic floating beam subjected to concentrated
load at midlength up to and beyond the formation of a three-plastic-hinge failure mechanism.
The beam was assumed to be perfectly elastic plastic with an ideal I-section. Furthermore, beam
deflections were considered small compared to the beam depth to comply with the simple beam
theory.
The application of the theory of beams on elastic foundation to wood joints has been
frequently reported in the literature. Most research employed the Winkler foundation model in an
attempt to fit a linear elastic load-slip relation. Yet, modern design methodologies shifted from
using linear elastic approximations to non-linear elastic-plastic analysis. This is in part attributed
to the fact that the application of the Winkler foundation model to wood gives moderate
predictions at best. Wood is not linear elastic when stressed to capacity and the analysis of joints
requires a look beyond the elastic limit. The European Yield theory (Johansen 1949), utilizes a
beam on a plastic foundation approach by holding onto the assumption that at capacity wood
crushing underneath the fastener is so advanced that the reaction force is uniformly distributed
along the fastener and the wood foundation is assumed to be perfectly plastic. While simplifying
the problem to a great extent and being capable of predicting joint capacity, the European Yield
Theory fails to relate capacity or any other loading state to joint slip. A more detailed discussion
of the Yield Theory is covered in a later section.
Foschi (1974) used a beam on an elastic-plastic foundation approach to derive a finite
element model capable of predicting the load-slip function of laterally loaded nails in wood.
Rather than using the linear elastic Winkler foundation, Foschi exploited a non-elastic foundation
model to account for crushing of the wood underneath the nail. The characteristics of the
foundation were expressed as
k x
p = (P0 + P1 x ) 1 e P0
(2.21)
Chapter 2
p
k
P1
P0
x
29
Unit
N
N/mm
N/mm
N
mm
For perfect yield (wood crushing at constant load), P1 = 0 and Equation (2.21) equals the
load slip relation reported more than four decades earlier by Teichmann and Brokmann (1930 and
1931). The constants k, P1 and P0 can be acquired from nonlinear least squares fitting of
experimental data obtained through embedment tests (Figure 2.15). Foschis model has been
used by many researchers (Dolan 1989, White 1995, Blass 1994, Frenette 1997) as input for
Load (p)
arctan P1
P0
arctan k
Slip (x)
approximated solutions with great accuracy of the most complex problems. Yet, the accuracy of
Chapter 2
30
numerical and analytical models rises and falls with the ability to quantify pertinent material
properties and their interactions which provide crucial information on material and connection
behavior that influence wood systems (Bodig et al. 1991).
There is an abundance of connection models reported in the literature. It is not the
purpose of this study to provide a complete description of all formulations attempting to predict
joint behavior.
Chapter 2
31
Europe
(B)
US
Embedment load
(A)
5% of bolt diameter
Slip
Chapter 2
32
predicted capacity
Fy
Fy
Fy
slip
slip
Fy
Fy
Fy
slip
slip
Fy
Fy
t
a
Fy
Fy
A1
Stress distribution Fe
ab
Fy
Fy
A2
ba
Condition of equilibrium:
A1 = A2
A3
Shear force
A4
Note that forces and moments due to asymmetry are assumed to be absorbed by the surrounding structure
and are not shown in the figures.
Chapter 2
33
Fy
F y = R = Fes t s d
(2.22)
tm
ts
F y = Fem t m d
d
Fy
F es
(2.23)
Chapter 2
34
F y = Fem b1 d = Fes b2 d
Fy
tm
(2.24)
If (Figure 2.19)
ts
M = Ra1 b1 + a1 + a1
2
1
1
Ra1 b1 + a1 Rb1 b1
2
2
(2.25)
then
Fy
F es
F em
b22
2 b12
M = Fem d a1
= Fes d
a 22
(2.26)
b2 = b1
F em
F es
a1 =
a1 a1 b1
b2 a 2 a 2
Rb 2
Ra 2
Ra 2
Ra1
M
Ra1
Rb1
Fem
Fes
t m b1
2
Fes
t s b1
t s b2 Fem
=
a2 =
F
2
2 es
Fem
(2.27)
(2.28)
(2.29)
Chapter 2
35
And substituting Equations (2.28) and (2.29) into Equation (2.25) and solving for b1 gives
tm
b1 =
F
1 + es
Fem
Fes
Fem
2
2
3
2
F
F
ts ts Fes ts
+ +
es + 2 es 1 +
Fem
Fem t m t m Fem t m
(2.30)
ts
1 +
t m
tm
F y = d Fem
F
1 + es
Fem
Fes
Fem
2
2
3
2
Fes t s t s Fes t s
Fes
1 +
+ 2
+ +
Fem
Fem t m t m Fem t m
t
1 + s
t m
(2.31)
Chapter 2
36
Fy
F y = Fem b1 d = Fes b2 d
tm
ts
(2.32)
M max = Rb1
b1
b
+ Rb 2 b1 + 2 +
2
2
(2.33)
a
3
Ra 2 b1 + b2 + 2 Ra 2 b1 + b2 + a 2
2
2
Fy
F es
or
b12
+ Fes d b2 b1 + 2 +
(2.34)
2
2
a
3
Fes d a 2 b1 + b2 + 2 b1 + b2 + a 2
2
2
M max = Fem d
F em
F es
b1
b2 a 2 a 2
b2 =
Rb 2
Ra 2
M max
Rb1
Ra 2
ts
F
2 es + 1
Fem
t s2
Fes
2
Fem
(2.35)
t s2
4 M max
+
+
2
F
2 es + 1 Fes d 2 Fes + 1
+ 1
F
Fem
em
Finally, substitution of b2 in Equation (2.32) with Equation (2.35) changes the expression of Fy to
Fy =
t s2
t s2
t s Fes d
4 M max
+
+
+ Fes d
2
F
F
Fes
2 es + 1
2 es + 1 Fes d 2 Fes + 1
+
1
F
Fem
Fem
F
em
em
(2.36)
Chapter 2
37
Fy
tm
F y = Fem b1 d = Fes b2 d
ts
(2.37)
2 M max = Rb1
b1
b
+ Rb 2 b1 + 2
2
2
(2.38)
or
Fy
2 M max
b12
b
= Fem d
+ Fes d b2 b1 + 2
2
2
(2.39)
F es
b1 =
F em
b1
2 M max
F
Fem d 1 + em
Fes
(2.40)
Hence
b2
F y = 2 Fem d
Rb 2
M max
M max
Rb1
Figure 2.21: Mode IV yield
M max
F
Fem d 1 + em
Fes
(2.41)
Chapter 2
Fy
Fem
Fes
d
Mmax
R
tm
ts
joint force
embedment strength of main member
embedment strength of side member
fastener diameter
maximum bending moment of fastener
force resultant
thickness of main member
thickness of side member
38
Unit
N
N/mm2
N/mm2
mm
Nmm
N
mm
mm
Chapter 2
39
Patton Mallory (1986) noted that because of the high variability of embedment strength,
yield strengths of failure modes depending more on embedment strength (i.e. fasteners with small
aspect ratios) generally show higher variability than strengths which are more dependent on
fastener bending. In addition, the European Yield Model does not account for oversized holes or
friction between joint members.